PETS 2016: Dataset and Challenge

Patino, Luis; Cane, Tom; Vallée, Annie; Ferryman, James

doi:10.1109/cvprw.2016.157

Cited by 44 publications

(27 citation statements)

References 1 publication

Supporting

Mentioning

Contrasting

Order By: Relevance

“…foreach frame at current time t detect all potential vessels V = {v k } N k=1 from the F t frame using M d53 foreach tracklet of vessel // Association gate generated from prior time Predict the kinematic track gate from MM filter from time t − 1 & adaptive association gate of appearance endfor MC data association using simulated annealing algorithm foreach pair of matched detection and tracklet tracklet update by MM filter endfor foreach detection vessel v k not associated with any tracklets in T We performed extensive experiments on SMD [1] and the PETS 2016 maritime dataset [6] to evaluate performance of the proposed M 3 C. The detector was pretrained on a Marvel vessel image dataset by the offline method [44].…”

Section: Algorithm 1: the Proposed M 3 C Tracking By Detectionmentioning

confidence: 99%

M3C: Multimodel-and-Multicue-Based Tracking by Detection of Surrounding Vessels in Maritime Environment for USV

et al. 2019

View full text Add to dashboard Cite

It is crucial for unmanned surface vessels (USVs) to detect and track surrounding vessels in real time to avoid collisions at sea. However, the harsh maritime environment poses great challenges to multitarget tracking (MTT). In this paper, a novel tracking by detection framework that integrates the multimodel and multicue (M3C) pipeline is proposed, which aims at improving the detection and tracking performance. Regarding the multimodel, we predicted the maneuver probability of a target vessel via the gated recurrent unit (GRU) model with an attention mechanism, and fused their respective outputs as the output of a kinematic filter. We developed a hybrid affinity model based on multi cues, such as the motion, appearance, and attitude of the ego vessel in the data association stage. By using the proposed ship re-identification approach, the tracker had the capability of appearance matching via metric learning. Experimental evaluation of two public maritime datasets showed that our method achieved state-of-the-art performance, not only in identity switches (IDS) but also in frame rates.

show abstract

Section: Algorithm 1: the Proposed M 3 C Tracking By Detectionmentioning

confidence: 99%

M3C: Multimodel-and-Multicue-Based Tracking by Detection of Surrounding Vessels in Maritime Environment for USV

et al. 2019

View full text Add to dashboard Cite

show abstract

“…We provide further details on both datasets in Section 4. Other popular examples of object tracking datasets include KITTI [67] and PETS [68].…”

Section: Related Workmentioning

confidence: 99%

Deep Affinity Network for Multiple Object Tracking

Sun

Akhtar

Song

et al. 2019

IEEE Trans. Pattern Anal. Mach. Intell.

239

140

View full text Add to dashboard Cite

Multiple Object Tracking (MOT) plays an important role in solving many fundamental problems in video analysis and computer vision. Most MOT methods employ two steps: Object Detection and Data Association. The first step detects objects of interest in every frame of a video, and the second establishes correspondence between the detected objects in different frames to obtain their tracks. Object detection has made tremendous progress in the last few years due to deep learning. However, data association for tracking still relies on hand crafted constraints such as appearance, motion, spatial proximity, grouping etc. to compute affinities between the objects in different frames. In this paper, we harness the power of deep learning for data association in tracking by jointly modeling object appearances and their affinities between different frames in an end-to-end fashion. The proposed Deep Affinity Network (DAN) learns compact, yet comprehensive features of pre-detected objects at several levels of abstraction, and performs exhaustive pairing permutations of those features in any two frames to infer object affinities. DAN also accounts for multiple objects appearing and disappearing between video frames. We exploit the resulting efficient affinity computations to associate objects in the current frame deep into the previous frames for reliable on-line tracking. Our technique is evaluated on popular multiple object tracking challenges MOT15, MOT17 and UA-DETRAC. Comprehensive benchmarking under twelve evaluation metrics demonstrates that our approach is among the best performing techniques on the leader board for these challenges. The open source implementation of our work is available at https://github.com/shijieS/SST.git.

show abstract

“…Labeled databases containing violence scenarios in surveillance video are scarce. We used here two important surveillance video databases publicly available online as BEHAVE [48] and ARENA [49]. As they contain a relatively small set of violent sequences, we decide to use both in this work.…”

Section: A Solution For Action Recognition In Low Hardware Resourcmentioning

confidence: 99%

A Sensor Network Approach for Violence Detection in Smart Cities Using Deep Learning

Baba

Gui²,

Cernazanu

et al. 2019

Sensors

View full text Add to dashboard Cite

Citizen safety in modern urban environments is an important aspect of life quality. Implementation of a smart city approach to video surveillance depends heavily on the capability of gathering and processing huge amounts of live urban data. Analyzing data from high bandwidth surveillance video streams provided by large size distributed sensor networks is particularly challenging. We propose here an efficient method for automatic violent behavior detection designed for video sensor networks. Known solutions to real-time violence detection are not suitable for implementation in a resource-constrained environment due to the high processing power requirements. Our algorithm achieves real-time processing on a Raspberry PI-embedded architecture. To ensure separation of temporal and spatial information processing we employ a computationally effective cascaded approach. It consists of a deep neural network followed by a time domain classifier. In contrast with current approaches, the deep neural network input is fed exclusively with motion vector features extracted directly from the MPEG encoded video stream. As proven by results, we achieve state-of-the-art performance, while running on a low computational resources embedded architecture.

show abstract

PETS 2016: Dataset and Challenge

Cited by 44 publications

References 1 publication

M3C: Multimodel-and-Multicue-Based Tracking by Detection of Surrounding Vessels in Maritime Environment for USV

M3C: Multimodel-and-Multicue-Based Tracking by Detection of Surrounding Vessels in Maritime Environment for USV

Deep Affinity Network for Multiple Object Tracking

A Sensor Network Approach for Violence Detection in Smart Cities Using Deep Learning

Contact Info

Product

Resources

About